Neighborhood Effects on Health in the Framingham Heart Study and the Role of Social Networks

Our objective in this project is to simultaneously assess the extent to which neighborhood of residence and embeddedness in defined (egocentric) social networks affect health. Since both networks and neighborhoods have effects on health, and since the presence of social network contacts might be interwoven with the effect of neighborhoods, our overarching objective is to ascertain the extent to which neighborhood effects depend on the propinquity of social network ties. We have four specific aims. First, we will embellish a longitudinal dataset, based on the Framingham Heart Study, describing 5,124 individuals (“egos”) and a social network of 12,630 people in which they are embedded (their possible “alters”), by adding detailed geocoding data for all these people measured roughly every four years from 1971 to 2007. As part of this aim, we will prepare detailed maps of where people and their social network contacts reside and visualize various neighborhood-level measures such as local wealth or crime. Second, we will describe the overlap between egos’ social network and neighborhood ties. We will examine the geographic distance between our egos and various kinds of alters and assess factors associated with this distance (e.g., the proximity of sisters versus brothers, the proximity of people to their friends). We will also assess which individuals have parochial (i.e., same neighborhood or same town) vs. cosmopolitan (i.e., outside) ties. Are there attributes of individuals, or of their micro-neighborhoods, that seem to foster maintenance of geographic propinquity? Third, using a hierarchical modeling framework, we will explore whether there are neighborhood effects on health outcomes in the FHS (the first time that the FHS has been used for this purpose). We will examine the dependence of individual health outcomes (i.e., mortality, cardiovascular disease, hypertension, obesity, depression, disability, and self-assessed health) on both individual attributes and on the attributes of neighborhoods (measured at the Census tract or block group level). We hypothesize that, in addition to substantial and significant variation at the individual level, there will be significant supra-individual variation in each of the outcomes that is attributable to supra-individual neighborhood context. Fourth, using a hierarchical modeling framework, we will evaluate the potential contribution of residing near one’s social network contacts in explaining neighborhood effects on individual health outcomes. We hypothesize that a substantial part of neighborhood effects on health can be attributed to the presence of social ties nearby. This work is policy-relevant since it will help localize the level at which such supra-individual health effects can arise and since it address the outcomes such as CVD, depression, and obesity, all of which are leading causes of morbidity in the elderly.